Drop a.out coredump support from gcore.
[dragonfly.git] / sys / kern / kern_fork.c
CommitLineData
984263bc
MD
1/*
2 * Copyright (c) 1982, 1986, 1989, 1991, 1993
3 * The Regents of the University of California. All rights reserved.
4 * (c) UNIX System Laboratories, Inc.
5 * All or some portions of this file are derived from material licensed
6 * to the University of California by American Telephone and Telegraph
7 * Co. or Unix System Laboratories, Inc. and are reproduced herein with
8 * the permission of UNIX System Laboratories, Inc.
9 *
10 * Redistribution and use in source and binary forms, with or without
11 * modification, are permitted provided that the following conditions
12 * are met:
13 * 1. Redistributions of source code must retain the above copyright
14 * notice, this list of conditions and the following disclaimer.
15 * 2. Redistributions in binary form must reproduce the above copyright
16 * notice, this list of conditions and the following disclaimer in the
17 * documentation and/or other materials provided with the distribution.
18 * 3. All advertising materials mentioning features or use of this software
19 * must display the following acknowledgement:
20 * This product includes software developed by the University of
21 * California, Berkeley and its contributors.
22 * 4. Neither the name of the University nor the names of its contributors
23 * may be used to endorse or promote products derived from this software
24 * without specific prior written permission.
25 *
26 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
27 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
28 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
29 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
30 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
31 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
32 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
33 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
34 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
35 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
36 * SUCH DAMAGE.
37 *
38 * @(#)kern_fork.c 8.6 (Berkeley) 4/8/94
5bc7cd8d 39 * $FreeBSD: src/sys/kern/kern_fork.c,v 1.72.2.14 2003/06/26 04:15:10 silby Exp $
cc1d939b 40 * $DragonFly: src/sys/kern/kern_fork.c,v 1.63 2007/02/18 16:17:09 corecode Exp $
984263bc
MD
41 */
42
43#include "opt_ktrace.h"
44
45#include <sys/param.h>
46#include <sys/systm.h>
47#include <sys/sysproto.h>
48#include <sys/filedesc.h>
49#include <sys/kernel.h>
50#include <sys/sysctl.h>
51#include <sys/malloc.h>
52#include <sys/proc.h>
53#include <sys/resourcevar.h>
54#include <sys/vnode.h>
55#include <sys/acct.h>
56#include <sys/ktrace.h>
dfc1fc13
EN
57#include <sys/unistd.h>
58#include <sys/jail.h>
59#include <sys/caps.h>
984263bc
MD
60
61#include <vm/vm.h>
62#include <sys/lock.h>
63#include <vm/pmap.h>
64#include <vm/vm_map.h>
65#include <vm/vm_extern.h>
66#include <vm/vm_zone.h>
67
68#include <sys/vmmeter.h>
69#include <sys/user.h>
e43a034f 70#include <sys/thread2.h>
984263bc
MD
71
72static MALLOC_DEFINE(M_ATFORK, "atfork", "atfork callback");
73
74/*
75 * These are the stuctures used to create a callout list for things to do
76 * when forking a process
77 */
78struct forklist {
79 forklist_fn function;
80 TAILQ_ENTRY(forklist) next;
81};
82
83TAILQ_HEAD(forklist_head, forklist);
84static struct forklist_head fork_list = TAILQ_HEAD_INITIALIZER(fork_list);
85
984263bc
MD
86int forksleep; /* Place for fork1() to sleep on. */
87
88/* ARGSUSED */
89int
753fd850 90sys_fork(struct fork_args *uap)
984263bc 91{
553ea3c8 92 struct lwp *lp = curthread->td_lwp;
984263bc 93 struct proc *p2;
41c20dac 94 int error;
984263bc 95
167e6ecb 96 error = fork1(lp, RFFDG | RFPROC | RFPGLOCK, &p2);
984263bc 97 if (error == 0) {
553ea3c8 98 start_forked_proc(lp, p2);
c7114eea
MD
99 uap->sysmsg_fds[0] = p2->p_pid;
100 uap->sysmsg_fds[1] = 0;
984263bc
MD
101 }
102 return error;
103}
104
105/* ARGSUSED */
106int
753fd850 107sys_vfork(struct vfork_args *uap)
984263bc 108{
553ea3c8 109 struct lwp *lp = curthread->td_lwp;
984263bc 110 struct proc *p2;
41c20dac 111 int error;
984263bc 112
167e6ecb 113 error = fork1(lp, RFFDG | RFPROC | RFPPWAIT | RFMEM | RFPGLOCK, &p2);
984263bc 114 if (error == 0) {
553ea3c8 115 start_forked_proc(lp, p2);
c7114eea
MD
116 uap->sysmsg_fds[0] = p2->p_pid;
117 uap->sysmsg_fds[1] = 0;
984263bc
MD
118 }
119 return error;
120}
121
f61c1ff1
MD
122/*
123 * Handle rforks. An rfork may (1) operate on the current process without
124 * creating a new, (2) create a new process that shared the current process's
125 * vmspace, signals, and/or descriptors, or (3) create a new process that does
126 * not share these things (normal fork).
127 *
128 * Note that we only call start_forked_proc() if a new process is actually
129 * created.
130 *
131 * rfork { int flags }
132 */
984263bc 133int
753fd850 134sys_rfork(struct rfork_args *uap)
984263bc 135{
553ea3c8 136 struct lwp *lp = curthread->td_lwp;
984263bc 137 struct proc *p2;
41c20dac 138 int error;
984263bc 139
6654fbcb
MD
140 if ((uap->flags & RFKERNELONLY) != 0)
141 return (EINVAL);
142
167e6ecb 143 error = fork1(lp, uap->flags | RFPGLOCK, &p2);
984263bc 144 if (error == 0) {
f61c1ff1 145 if (p2)
553ea3c8 146 start_forked_proc(lp, p2);
c7114eea
MD
147 uap->sysmsg_fds[0] = p2 ? p2->p_pid : 0;
148 uap->sysmsg_fds[1] = 0;
984263bc
MD
149 }
150 return error;
151}
152
153
154int nprocs = 1; /* process 0 */
984263bc
MD
155
156int
553ea3c8 157fork1(struct lwp *lp1, int flags, struct proc **procp)
984263bc 158{
553ea3c8 159 struct proc *p1 = lp1->lwp_proc;
984263bc 160 struct proc *p2, *pptr;
167e6ecb 161 struct pgrp *pgrp;
553ea3c8 162 struct lwp *lp2;
984263bc 163 uid_t uid;
167e6ecb 164 int ok, error;
51e64ff2 165 static int curfail = 0;
5bc7cd8d 166 static struct timeval lastfail;
984263bc
MD
167 struct forklist *ep;
168 struct filedesc_to_leader *fdtol;
169
170 if ((flags & (RFFDG|RFCFDG)) == (RFFDG|RFCFDG))
171 return (EINVAL);
172
173 /*
174 * Here we don't create a new process, but we divorce
175 * certain parts of a process from itself.
176 */
177 if ((flags & RFPROC) == 0) {
178
bb3cd951 179 vm_fork(lp1, 0, flags);
984263bc
MD
180
181 /*
182 * Close all file descriptors.
183 */
184 if (flags & RFCFDG) {
185 struct filedesc *fdtmp;
186 fdtmp = fdinit(p1);
187 fdfree(p1);
188 p1->p_fd = fdtmp;
189 }
190
191 /*
192 * Unshare file descriptors (from parent.)
193 */
194 if (flags & RFFDG) {
195 if (p1->p_fd->fd_refcnt > 1) {
196 struct filedesc *newfd;
197 newfd = fdcopy(p1);
198 fdfree(p1);
199 p1->p_fd = newfd;
200 }
201 }
202 *procp = NULL;
203 return (0);
204 }
205
167e6ecb
MD
206 /*
207 * Interlock against process group signal delivery. If signals
208 * are pending after the interlock is obtained we have to restart
209 * the system call to process the signals. If we don't the child
210 * can miss a pgsignal (such as ^C) sent during the fork.
211 *
212 * We can't use CURSIG() here because it will process any STOPs
213 * and cause the process group lock to be held indefinitely. If
214 * a STOP occurs, the fork will be restarted after the CONT.
215 */
216 error = 0;
217 pgrp = NULL;
218 if ((flags & RFPGLOCK) && (pgrp = p1->p_pgrp) != NULL) {
219 lockmgr(&pgrp->pg_lock, LK_SHARED);
08f2f1bb 220 if (CURSIGNB(lp1)) {
167e6ecb
MD
221 error = ERESTART;
222 goto done;
223 }
224 }
225
984263bc
MD
226 /*
227 * Although process entries are dynamically created, we still keep
228 * a global limit on the maximum number we will create. Don't allow
229 * a nonprivileged user to use the last ten processes; don't let root
230 * exceed the limit. The variable nprocs is the current number of
231 * processes, maxproc is the limit.
232 */
41c20dac 233 uid = p1->p_ucred->cr_ruid;
984263bc 234 if ((nprocs >= maxproc - 10 && uid != 0) || nprocs >= maxproc) {
5bc7cd8d 235 if (ppsratecheck(&lastfail, &curfail, 1))
6ea70f76 236 kprintf("maxproc limit exceeded by uid %d, please "
5bc7cd8d 237 "see tuning(7) and login.conf(5).\n", uid);
377d4740 238 tsleep(&forksleep, 0, "fork", hz / 2);
167e6ecb
MD
239 error = EAGAIN;
240 goto done;
984263bc
MD
241 }
242 /*
243 * Increment the nprocs resource before blocking can occur. There
244 * are hard-limits as to the number of processes that can run.
245 */
246 nprocs++;
247
248 /*
249 * Increment the count of procs running with this uid. Don't allow
250 * a nonprivileged user to exceed their current limit.
251 */
41c20dac 252 ok = chgproccnt(p1->p_ucred->cr_ruidinfo, 1,
984263bc
MD
253 (uid != 0) ? p1->p_rlimit[RLIMIT_NPROC].rlim_cur : 0);
254 if (!ok) {
255 /*
256 * Back out the process count
257 */
258 nprocs--;
5bc7cd8d 259 if (ppsratecheck(&lastfail, &curfail, 1))
6ea70f76 260 kprintf("maxproc limit exceeded by uid %d, please "
5bc7cd8d 261 "see tuning(7) and login.conf(5).\n", uid);
377d4740 262 tsleep(&forksleep, 0, "fork", hz / 2);
167e6ecb
MD
263 error = EAGAIN;
264 goto done;
984263bc
MD
265 }
266
267 /* Allocate new proc. */
51e64ff2 268 p2 = zalloc(proc_zone);
08f2f1bb 269 lp2 = zalloc(lwp_zone);
984263bc
MD
270
271 /*
ef09c3ed 272 * Setup linkage for kernel based threading XXX lwp
984263bc 273 */
cb74210d 274 if (flags & RFTHREAD) {
51e64ff2
MD
275 p2->p_peers = p1->p_peers;
276 p1->p_peers = p2;
277 p2->p_leader = p1->p_leader;
984263bc 278 } else {
51e64ff2
MD
279 p2->p_peers = NULL;
280 p2->p_leader = p2;
984263bc
MD
281 }
282
51e64ff2
MD
283 p2->p_wakeup = 0;
284 p2->p_vmspace = NULL;
285 p2->p_numposixlocks = 0;
286 p2->p_emuldata = NULL;
51e64ff2 287 LIST_INIT(&p2->p_lwps);
ef09c3ed
SS
288
289 /* XXX lwp */
51e64ff2 290 lp2->lwp_proc = p2;
553ea3c8 291 lp2->lwp_tid = 0;
51e64ff2
MD
292 LIST_INSERT_HEAD(&p2->p_lwps, lp2, lwp_list);
293 p2->p_nthreads = 1;
294 p2->p_nstopped = 0;
295 p2->p_lasttid = 0;
984263bc
MD
296
297 /*
51e64ff2
MD
298 * Setting the state to SIDL protects the partially initialized
299 * process once it starts getting hooked into the rest of the system.
984263bc 300 */
51e64ff2 301 p2->p_stat = SIDL;
164b8401 302 lp2->lwp_stat = LSRUN; /* XXX use other state? start_forked_proc() handles this*/
51e64ff2 303 proc_add_allproc(p2);
984263bc
MD
304
305 /*
306 * Make a proc table entry for the new process.
307 * Start by zeroing the section of proc that is zero-initialized,
308 * then copy the section that is copied directly from the parent.
309 */
310 bzero(&p2->p_startzero,
311 (unsigned) ((caddr_t)&p2->p_endzero - (caddr_t)&p2->p_startzero));
553ea3c8
SS
312 bzero(&lp2->lwp_startzero,
313 (unsigned) ((caddr_t)&lp2->lwp_endzero -
314 (caddr_t)&lp2->lwp_startzero));
984263bc
MD
315 bcopy(&p1->p_startcopy, &p2->p_startcopy,
316 (unsigned) ((caddr_t)&p2->p_endcopy - (caddr_t)&p2->p_startcopy));
08f2f1bb 317 bcopy(&lp1->lwp_startcopy, &lp2->lwp_startcopy,
553ea3c8
SS
318 (unsigned) ((caddr_t)&lp2->lwp_endcopy -
319 (caddr_t)&lp2->lwp_startcopy));
984263bc
MD
320
321 p2->p_aioinfo = NULL;
322
323 /*
324 * Duplicate sub-structures as needed.
325 * Increase reference counts on shared objects.
326 * The p_stats and p_sigacts substructs are set in vm_fork.
3a72e617 327 * p_lock is in the copy area and must be cleared.
984263bc 328 */
344ad853 329 p2->p_flag = 0;
3a72e617 330 p2->p_lock = 0;
cc1d939b 331 lp2->lwp_lock = 0;
984263bc
MD
332 if (p1->p_flag & P_PROFIL)
333 startprofclock(p2);
dadab5e9 334 p2->p_ucred = crhold(p1->p_ucred);
984263bc 335
b40e316c 336 if (jailed(p2->p_ucred))
984263bc 337 p2->p_flag |= P_JAILED;
984263bc
MD
338
339 if (p2->p_args)
340 p2->p_args->ar_ref++;
341
342 if (flags & RFSIGSHARE) {
343 p2->p_procsig = p1->p_procsig;
344 p2->p_procsig->ps_refcnt++;
345 if (p1->p_sigacts == &p1->p_addr->u_sigacts) {
346 struct sigacts *newsigacts;
984263bc
MD
347
348 /* Create the shared sigacts structure */
349 MALLOC(newsigacts, struct sigacts *,
350 sizeof(struct sigacts), M_SUBPROC, M_WAITOK);
e43a034f 351 crit_enter();
984263bc
MD
352 /*
353 * Set p_sigacts to the new shared structure.
354 * Note that this is updating p1->p_sigacts at the
355 * same time, since p_sigacts is just a pointer to
356 * the shared p_procsig->ps_sigacts.
357 */
358 p2->p_sigacts = newsigacts;
359 bcopy(&p1->p_addr->u_sigacts, p2->p_sigacts,
360 sizeof(*p2->p_sigacts));
361 *p2->p_sigacts = p1->p_addr->u_sigacts;
e43a034f 362 crit_exit();
984263bc
MD
363 }
364 } else {
365 MALLOC(p2->p_procsig, struct procsig *, sizeof(struct procsig),
366 M_SUBPROC, M_WAITOK);
367 bcopy(p1->p_procsig, p2->p_procsig, sizeof(*p2->p_procsig));
368 p2->p_procsig->ps_refcnt = 1;
369 p2->p_sigacts = NULL; /* finished in vm_fork() */
370 }
371 if (flags & RFLINUXTHPN)
372 p2->p_sigparent = SIGUSR1;
373 else
374 p2->p_sigparent = SIGCHLD;
375
376 /* bump references to the text vnode (for procfs) */
377 p2->p_textvp = p1->p_textvp;
378 if (p2->p_textvp)
597aea93 379 vref(p2->p_textvp);
984263bc 380
0daa37a5
MD
381 /*
382 * Handle file descriptors
383 */
984263bc
MD
384 if (flags & RFCFDG) {
385 p2->p_fd = fdinit(p1);
386 fdtol = NULL;
387 } else if (flags & RFFDG) {
388 p2->p_fd = fdcopy(p1);
389 fdtol = NULL;
390 } else {
391 p2->p_fd = fdshare(p1);
392 if (p1->p_fdtol == NULL)
393 p1->p_fdtol =
394 filedesc_to_leader_alloc(NULL,
395 p1->p_leader);
396 if ((flags & RFTHREAD) != 0) {
397 /*
398 * Shared file descriptor table and
399 * shared process leaders.
400 */
401 fdtol = p1->p_fdtol;
402 fdtol->fdl_refcount++;
403 } else {
404 /*
405 * Shared file descriptor table, and
406 * different process leaders
407 */
98a7f915 408 fdtol = filedesc_to_leader_alloc(p1->p_fdtol, p2);
984263bc
MD
409 }
410 }
411 p2->p_fdtol = fdtol;
c0b8a06d 412 p2->p_limit = plimit_fork(p1->p_limit);
984263bc
MD
413
414 /*
415 * Preserve some more flags in subprocess. P_PROFIL has already
416 * been preserved.
417 */
08f2f1bb
SS
418 p2->p_flag |= p1->p_flag & P_SUGID;
419 lp2->lwp_flag |= lp1->lwp_flag & LWP_ALTSTACK;
984263bc
MD
420 if (p1->p_session->s_ttyvp != NULL && p1->p_flag & P_CONTROLT)
421 p2->p_flag |= P_CONTROLT;
422 if (flags & RFPPWAIT)
423 p2->p_flag |= P_PPWAIT;
424
0daa37a5
MD
425 /*
426 * Inherit the virtual kernel structure (allows a virtual kernel
427 * to fork to simulate multiple cpus).
428 */
4a22e893
MD
429 p2->p_vkernel = NULL;
430 if (p1->p_vkernel)
431 vkernel_inherit(p1, p2);
0daa37a5 432
5fd012e0
MD
433 /*
434 * Once we are on a pglist we may receive signals. XXX we might
435 * race a ^C being sent to the process group by not receiving it
436 * at all prior to this line.
437 */
984263bc
MD
438 LIST_INSERT_AFTER(p1, p2, p_pglist);
439
440 /*
441 * Attach the new process to its parent.
442 *
443 * If RFNOWAIT is set, the newly created process becomes a child
444 * of init. This effectively disassociates the child from the
445 * parent.
446 */
447 if (flags & RFNOWAIT)
448 pptr = initproc;
449 else
450 pptr = p1;
451 p2->p_pptr = pptr;
452 LIST_INSERT_HEAD(&pptr->p_children, p2, p_sibling);
453 LIST_INIT(&p2->p_children);
98a7f915 454 varsymset_init(&p2->p_varsymset, &p1->p_varsymset);
8fbf9130 455 callout_init(&p2->p_ithandle);
984263bc
MD
456
457#ifdef KTRACE
458 /*
459 * Copy traceflag and tracefile if enabled. If not inherited,
460 * these were zeroed above but we still could have a trace race
29f58392 461 * so make sure p2's p_tracenode is NULL.
984263bc 462 */
29f58392 463 if ((p1->p_traceflag & KTRFAC_INHERIT) && p2->p_tracenode == NULL) {
984263bc 464 p2->p_traceflag = p1->p_traceflag;
29f58392 465 p2->p_tracenode = ktrinherit(p1->p_tracenode);
984263bc
MD
466 }
467#endif
468
469 /*
352f5709 470 * Inherit the scheduler and initialize scheduler-related fields.
dcc99b62
MD
471 * Set cpbase to the last timeout that occured (not the upcoming
472 * timeout).
50017724
MD
473 *
474 * A critical section is required since a timer IPI can update
475 * scheduler specific data.
a77ac49d 476 */
50017724 477 crit_enter();
a77ac49d 478 p2->p_usched = p1->p_usched;
553ea3c8 479 lp2->lwp_cpbase = mycpu->gd_schedclock.time -
dcc99b62 480 mycpu->gd_schedclock.periodic;
08f2f1bb 481 p2->p_usched->heuristic_forking(lp1, lp2);
50017724 482 crit_exit();
a77ac49d 483
984263bc
MD
484 /*
485 * This begins the section where we must prevent the parent
486 * from being swapped.
487 */
488 PHOLD(p1);
489
490 /*
491 * Finish creating the child process. It will return via a different
492 * execution path later. (ie: directly into user mode)
493 */
bb3cd951 494 vm_fork(lp1, p2, flags);
08f2f1bb 495 caps_fork(lp1->lwp_thread, lp2->lwp_thread, flags);
984263bc
MD
496
497 if (flags == (RFFDG | RFPROC)) {
12e4aaff
MD
498 mycpu->gd_cnt.v_forks++;
499 mycpu->gd_cnt.v_forkpages += p2->p_vmspace->vm_dsize + p2->p_vmspace->vm_ssize;
984263bc 500 } else if (flags == (RFFDG | RFPROC | RFPPWAIT | RFMEM)) {
12e4aaff
MD
501 mycpu->gd_cnt.v_vforks++;
502 mycpu->gd_cnt.v_vforkpages += p2->p_vmspace->vm_dsize + p2->p_vmspace->vm_ssize;
984263bc 503 } else if (p1 == &proc0) {
12e4aaff
MD
504 mycpu->gd_cnt.v_kthreads++;
505 mycpu->gd_cnt.v_kthreadpages += p2->p_vmspace->vm_dsize + p2->p_vmspace->vm_ssize;
984263bc 506 } else {
12e4aaff
MD
507 mycpu->gd_cnt.v_rforks++;
508 mycpu->gd_cnt.v_rforkpages += p2->p_vmspace->vm_dsize + p2->p_vmspace->vm_ssize;
984263bc
MD
509 }
510
511 /*
512 * Both processes are set up, now check if any loadable modules want
513 * to adjust anything.
514 * What if they have an error? XXX
515 */
516 TAILQ_FOREACH(ep, &fork_list, next) {
517 (*ep->function)(p1, p2, flags);
518 }
519
520 /*
a77ac49d
MD
521 * Set the start time. Note that the process is not runnable. The
522 * caller is responsible for making it runnable.
984263bc 523 */
d9fa5f67 524 microtime(&p2->p_start);
984263bc 525 p2->p_acflag = AFORK;
984263bc
MD
526
527 /*
528 * tell any interested parties about the new process
529 */
530 KNOTE(&p1->p_klist, NOTE_FORK | p2->p_pid);
531
984263bc
MD
532 /*
533 * Return child proc pointer to parent.
534 */
535 *procp = p2;
167e6ecb
MD
536done:
537 if (pgrp)
538 lockmgr(&pgrp->pg_lock, LK_RELEASE);
539 return (error);
984263bc
MD
540}
541
542/*
543 * The next two functionms are general routines to handle adding/deleting
544 * items on the fork callout list.
545 *
546 * at_fork():
547 * Take the arguments given and put them onto the fork callout list,
548 * However first make sure that it's not already there.
549 * Returns 0 on success or a standard error number.
550 */
984263bc 551int
303c76d5 552at_fork(forklist_fn function)
984263bc
MD
553{
554 struct forklist *ep;
555
556#ifdef INVARIANTS
557 /* let the programmer know if he's been stupid */
303c76d5 558 if (rm_at_fork(function)) {
6ea70f76 559 kprintf("WARNING: fork callout entry (%p) already present\n",
984263bc 560 function);
303c76d5 561 }
984263bc 562#endif
efda3bd0 563 ep = kmalloc(sizeof(*ep), M_ATFORK, M_WAITOK|M_ZERO);
984263bc
MD
564 ep->function = function;
565 TAILQ_INSERT_TAIL(&fork_list, ep, next);
566 return (0);
567}
568
569/*
570 * Scan the exit callout list for the given item and remove it..
571 * Returns the number of items removed (0 or 1)
572 */
984263bc 573int
303c76d5 574rm_at_fork(forklist_fn function)
984263bc
MD
575{
576 struct forklist *ep;
577
578 TAILQ_FOREACH(ep, &fork_list, next) {
579 if (ep->function == function) {
580 TAILQ_REMOVE(&fork_list, ep, next);
efda3bd0 581 kfree(ep, M_ATFORK);
984263bc
MD
582 return(1);
583 }
584 }
585 return (0);
586}
7d0bac62
MD
587
588/*
589 * Add a forked process to the run queue after any remaining setup, such
590 * as setting the fork handler, has been completed.
591 */
7d0bac62 592void
553ea3c8 593start_forked_proc(struct lwp *lp1, struct proc *p2)
7d0bac62 594{
08f2f1bb 595 struct lwp *lp2 = ONLY_LWP_IN_PROC(p2);
553ea3c8 596
7d0bac62 597 /*
26a0694b
MD
598 * Move from SIDL to RUN queue, and activate the process's thread.
599 * Activation of the thread effectively makes the process "a"
600 * current process, so we do not setrunqueue().
8ec60c3f
MD
601 *
602 * YYY setrunqueue works here but we should clean up the trampoline
603 * code so we just schedule the LWKT thread and let the trampoline
604 * deal with the userland scheduler on return to userland.
7d0bac62 605 */
553ea3c8 606 KASSERT(p2->p_stat == SIDL,
7d0bac62 607 ("cannot start forked process, bad status: %p", p2));
553ea3c8 608 p2->p_usched->resetpriority(lp2);
e43a034f 609 crit_enter();
164b8401
SS
610 p2->p_stat = SACTIVE;
611 lp2->lwp_stat = LSRUN;
553ea3c8 612 p2->p_usched->setrunqueue(lp2);
e43a034f 613 crit_exit();
7d0bac62
MD
614
615 /*
616 * Now can be swapped.
617 */
553ea3c8 618 PRELE(lp1->lwp_proc);
7d0bac62
MD
619
620 /*
621 * Preserve synchronization semantics of vfork. If waiting for
622 * child to exec or exit, set P_PPWAIT on child, and sleep on our
623 * proc (in case of exit).
624 */
625 while (p2->p_flag & P_PPWAIT)
553ea3c8 626 tsleep(lp1->lwp_proc, 0, "ppwait", 0);
7d0bac62 627}